How BAMF works
One question. One probability. A time-stamped learning record.
1. Forecast a probability
State the chance before the outcome. The exact crowd forecast stays hidden by default.
2. Choose your privacy
Keep it for yourself or publish it deliberately. Private is always the default.
3. Resolve from a named source
Every question carries a cutoff, source, rule, and invalidation path.
4. Learn from a series, not one result
Calibration asks whether events forecast near 70% happen about 70% of the time.
Your question. Your forecast.
Create a verifiable Yes or No question, define its cutoff and resolution source, then make your call.